On 01/05/2012 10:27 AM, Ulrich Windl wrote:
> Hi!
>
> Looking into the Filesystem RA of SLES11 SP1, I'm wondering:
>
> Previously I found out that the RA happily mounts a filesystem with
> errors. Now I found out that the RA can check the filesystems
> "automatically" (default). Looking how it works makes me woner:
>
> is_fsck_needed() { case $OCF_RESKEY_run_fsck in force) true;; no)
> false;; ""|auto) case $FSTYPE in
> ext4|ext4dev|ext3|reiserfs|reiser4|nss|xfs|jfs|vfat|fat|nfs\
> |cifs|smbfs|ocfs2|gfs2|none|lustre) false;; *) true;; esac;; *)
> ocf_log warn "Invalid parameter value for fsck: '$OCF_RESKEY_ru\
> n_fsck'; setting to 'auto'" OCF_RESKEY_run_fsck="auto"
> is_fsck_needed;; esac }
>
> So a filesystem is checked if it's "force-d", it's not tchecked if
> you say "no", and in the "auto" case it's not checked if it's a
> well-known filesystem! ???
>
> So you are saying those filesystems listed never have errors?
>
> I'd agree on "nfs", "cifs", "smbfs" and "none", because it does not
> make sense to check those, but for the ext3 case I'm wondering quite
> a lot.
>
> I'm surprised about those design decisions. From tune2fs: You should
> strongly consider the consequences of disabling
> mount-count-dependent checking entirely. Bad disk drives,
> cables, memory, and kernel bugs could all corrupt a filesystem
> without marking the filesystem dirty or in error. If you are
> using journaling on your filesystem, your filesystem will never be
> marked dirty, so it will not normally be checked. A filesys- tem
> error detected by the kernel will still force an fsck on the next
> reboot, but it may already be too late to prevent data loss at that
> point.
>
> Note that with the Filesystem RA, the filesystem will _never_ be
> checked (ignoring the settings)!
>
> (Recently I got a report from a consultant that stated that it's not
> possible with pacemaker to implement a "light out" functionality
> (probably meaning that the cluster can reliably manage system
> faults). Looking at various implementations details and omissions, I
> must agree that he's right. Probably it's time to update the
> documentation.)
I think you should check the archives about your question - fsck has
been discussed before.
> http://lists.community.tummy.com/pipermail/linux-ha-dev/2010-June/017475.html
With recent e2fsprogs versions the agent could refuse to mount at all if
errors are detected:
> http://comments.gmane.org/gmane.comp.file-systems.ext4/21765
However, I'm not going to write a patch for the filesystem agent, as I'm
right now not working with pacemaker anyway and as I think the generic
filesystem agent is far too complex for a shell script anyway.
Cheers,
Bernd
_______________________________________________
Linux-HA mailing list
[email protected]
http://lists.linux-ha.org/mailman/listinfo/linux-ha
See also: http://linux-ha.org/ReportingProblems
